This revised R34 proposal addresses one of the most significant problems in child mental health services: poor adherence to evidence-based treatment. Poor adherence leads to unsatisfactory patient outcomes, wastes billions of dollars in avoidable healthcare expenditures, and can have life-long consequences, affecting children's future physical health, social, and economic well-being. Traditionally the focus has been on improving service systems but this has failed to engage and retain individuals in treatment. Alternative approaches to overcome poor adherence are needed. The PI's previous NIMH-funded study established the proof-of-concept linking caregiver approaches to care with differential propensities for treatment adherence. The primary innovation in this proposal is the use of discrete choice experiments and latent class analysis to identify segments of caregivers with similar preferences and to identify physicians' awareness of these beliefs as they relate to adherence. This study specifically aims to 1) identify the treatment attributes caregivers' most value for their child's ADHD and any changes in valuation over a 6-month course of care; 2) evaluate the concordance between caregivers' beliefs about treatment and physicians' views of caregivers' beliefs about treatment at the initial visit; and 3) determine if caregiver's treatment preferences and caregiver-physician concordance on beliefs about treatment predict adherence to clinic attendance and medication use. A sample of 50 mental health physicians and 300 caregivers of the physicians' new patients 6-12 years old and diagnosed with ADHD will be recruited from child-serving mental health settings across the state of Maryland. Caregivers' beliefs and preferences for treatment will be measured at baseline, 3, and 6 months using a discrete choice experiment. Physicians complete a baseline survey of their predictions of the preferences of each caregiver and of their practice and professional characteristics. The discrete choice experiment data will be analyzed using mixed effects logistic regression models. Latent class analysis will be used to segment the caregiver sample into preference subgroups. Caregiver-physician concordance will also be assessed. Adherence is measured by clinic attendance and medication use for ADHD. The impact of caregiver preferences and caregiver-physician concordance on adherence, adjusting for child symptoms, provider alliance, and parental burden, will be analyzed using models for longitudinal data. Using treatment preferences to predict adherence is an extremely innovative approach to a serious and complex public health problem. The multidisciplinary team and the strong connection to community providers are an excellent environment for the proposed research. This work will promote knowledge transfer for sustainable child mental health services. The potential for generalizing this to primary care settings and to other child mental health conditions will move the field towards better outcomes for children and their families.